SPB Git

Repository Public Research preview

summary.md 2.7 KB

system n chrf edit_sim rougeL exact_match number_recall placeholders_exact
B0_identity 502 71.239 [69.692, 72.653] 0.666 [0.645, 0.686] 0.690 [0.670, 0.708] 0.000 [0.000, 0.000] 1.000 [1.000, 1.000] 1.000 [1.000, 1.000]
B1_generic_paraphrase 502 68.415 [66.896, 69.992] 0.561 [0.545, 0.579] 0.612 [0.593, 0.632] 0.000 [0.000, 0.000] 1.000 [1.000, 1.000] 1.000 [1.000, 1.000]
B2_t5-small 502 68.849 [67.661, 69.944] 0.638 [0.621, 0.657] 0.671 [0.654, 0.687] 0.000 [0.000, 0.000] 0.945 [0.919, 0.969] 0.691 [0.585, 0.787]
htr-denoise-only 502 58.059 [55.208, 60.869] 0.456 [0.421, 0.491] 0.493 [0.459, 0.526] 0.000 [0.000, 0.000] 0.856 [0.806, 0.902] 0.741 [0.649, 0.824]
htr-m1-tiny-denoiseinit 502 63.605 [62.198, 64.861] 0.578 [0.558, 0.595] 0.600 [0.579, 0.617] 0.000 [0.000, 0.000] 0.931 [0.912, 0.950] 0.884 [0.830, 0.938]
htr-m1-tiny-mediumdenoiseinit 502 67.913 [66.433, 69.378] 0.630 [0.611, 0.649] 0.656 [0.637, 0.674] 0.000 [0.000, 0.000] 0.962 [0.942, 0.976] 0.932 [0.892, 0.977]
htr-m1-tiny 502 32.864 [30.871, 34.474] 0.287 [0.274, 0.299] 0.241 [0.224, 0.255] 0.000 [0.000, 0.000] 0.571 [0.503, 0.659] 0.633 [0.505, 0.742]
htr-pilot_m1_tiny_denoiseinit_s2 502 64.122 [62.539, 65.454] 0.582 [0.562, 0.600] 0.603 [0.581, 0.622] 0.000 [0.000, 0.000] 0.936 [0.909, 0.959] 0.878 [0.819, 0.936]
htr-pilot_m3light_tiny_denoiseinit 502 63.999 [62.562, 65.346] 0.584 [0.563, 0.605] 0.606 [0.584, 0.625] 0.000 [0.000, 0.000] 0.941 [0.923, 0.964] 0.878 [0.830, 0.931]
htr-pilot_m3tokenonly_tiny_denoiseinit 502 64.058 [62.248, 65.585] 0.584 [0.561, 0.605] 0.606 [0.582, 0.627] 0.000 [0.000, 0.000] 0.952 [0.935, 0.970] 0.890 [0.850, 0.935]

Stylometric W1 (mean over features, standardised; lower = closer to originals):

  • rewrite vs original: 0.170
  • B0_identity vs original: 0.170 · probe P(original) mean 0.44
  • B1_generic_paraphrase vs original: 0.281 · probe P(original) mean 0.44
  • B2_t5-small vs original: 0.112 · probe P(original) mean 0.54
  • htr-denoise-only vs original: 0.215 · probe P(original) mean 0.53
  • htr-m1-tiny-denoiseinit vs original: 0.150 · probe P(original) mean 0.55
  • htr-m1-tiny-mediumdenoiseinit vs original: 0.137 · probe P(original) mean 0.55
  • htr-m1-tiny vs original: 1.699 · probe P(original) mean 0.73
  • htr-pilot_m1_tiny_denoiseinit_s2 vs original: 0.138 · probe P(original) mean 0.54
  • htr-pilot_m3light_tiny_denoiseinit vs original: 0.155 · probe P(original) mean 0.53
  • htr-pilot_m3tokenonly_tiny_denoiseinit vs original: 0.159 · probe P(original) mean 0.52
Files main
Menu

Theme